Download PDF

ASSESSING THE IMPACT OF CLIMATE CHANGE ON AGRICULTURE IN GERMANY
- A RICARDIAN ANALYSIS
JOACHIM AURBACHER, CHRISTIAN LIPPERT, and TATJANA KRIMLY
Institute of Farm Management (410A), Universität Hohenheim, Germany,
[email protected]
Contributed paper at the
IATRC Public Trade Policy Research and Analysis Symposium
‘Climate Change in World Agriculture: Mitigation, Adaptation, Trade and Food Security’
Universität Hohenheim, Stuttgart, Germany, June 27-29, 2010
Copyright 2010 by Joachim Aurbacher, Christian Lippert, and Tatjana Krimly. All rights
reserved. Readers may make verbatim copies of this document for non-commercial purposes
by any means, provided that this copyright notice appears on all such copies.
Abstract
To assess the impact of climate change on agriculture in German districts, a Ricardian
analysis accounting for spatial autocorrelation is used. The analysis relies on land rental prices
from agricultural statistics and climate data from the network of German weather observation
stations. The results of the cross-sectional analysis show a significant correlation between
land rents and increasing mean temperatures, as well as (except for East Germany) a
significant increase of land rents along with declining spring precipitation. Next, climate data
from the regional climate model REMO are used to calculate local land rent changes under
different IPCC scenarios for a time horizon between 2011 and 2040. The overall result is an
increase in land rents which relates to 5-6 % of net German agricultural income. In West
Germany the increase in land rents rises from the north to south, because of the higher
temperature increases in the South of Germany. East Germany shows a greater increase in
land rents than West Germany, due to the presently more unfavourable climatic water balance
which profits from the increase in spring precipitation. However, agricultural income in
Germany might decrease in the long run, when the changes in the climatic conditions will be
more severe than those simulated in this study. We plan to extend and improve the research
in this field. Different approaches to consider soil qualities at district level shall be compared.
In order to understand farmers’ adaptation to different climatic conditions multinomial logit
models will be used to investigate how German land use structures depend on the climate.
Furthermore, a typology of related transaction costs shall be devised. This will allow the
identification of those regions where the adaptation needs and the corresponding costs are
likely to be relatively high.
1
Introduction
In the upcoming years for Germany an increase in temperatures and a modified distribution of
precipitation is being expected. Likely a higher precipitation in winter will be attended by
drier summers and an overall warming.
Figure 1 shows the shifts of the mean annual temperatures and the spring precipitation in the
German districts for the IPCC scenario A1B on base of the REMO-Modell (MPI on behalf of
the Umweltbundesamt, 2006). This scenario assumes a world with high economic growth and
a population growth until mid-century as well as a balanced use of fossil and non-fossil
energy sources. In addition, a larger variability of the weather events can be expected: storms
and extreme summer droughts are likely to occur more frequently (cf. Zebisch et al., 2005, p.
38f).
In order to assess the effects of such a climate change on German agriculture we performed a
Ricardian analysis relying on the land rental price data represented in Figure 2 which had
been communicated by Statistisches Landesamt Baden-Württemberg (2007).
2
Figure 1: Expected changes in (a) annual mean temperature and (b) spring precipitation
(March - June) for IPCC scenario A1B according to REMO; Changes 2011-2041
against 1961-1990
Source: Calculations on the basis of DWD (2007) und MPI on behalf of the
Umweltbundesamt (2006); depiction from Lippert et al. (2009)
Figure 2: Land rental prices per ha UAA in the German districts in 1999
Source: Statistisches Landesamt Baden-Württemberg (2007); depiction from
Lippert et al. (2009)
3
2
Method
Equations 1 and 2 outline the cross-sectional statistical approach applied (Anselin 1988: 34ff.;
LeSage 1999: 52f.) which accounts for spatial error dependence (due to spatially correlated
omitted explaining variables), since we could not obtain data for all possibly relevant
variables:
r 
X
 u
(1)
u  Wu  
with
ε
(2)
~ N(0,σ2I),
where
r
= n x 1 vector of reported average farmland rental prices for the different districts
(Landkreise) i (i = 1, …, n);
X = n x (1+k) matrix containing a set of observations for k explanatory climate and nonclimate variables;
W = standardised first order contiguity matrix;
I
= n x n identity matrix;
u = n x 1 vector of the spatially correlated residuals;
ε
= n x 1 vector of normally distributed errors (mean = 0, variance = σ2).
The vector of regression coefficients β and the parameter λ were estimated. Further details of
our Ricardian analysis along with an overview on recent studies in this field are given in
Lippert et al. (2009) (for the basic principles of the method cf. the seminal paper by
Mendelsohn et al., 1994).
3
Results of the Ricardian Analysis
Explaining land rental prices represented in Figure 2 by spatially interpolated climate data
from the weather observation stations of the German Weather Service (DWD, 2007) we
obtained the results outlined in Table 1 (cf. Lippert et al., 2009):
4
Table 1: Ricardian regression models for German land rental prices a)
Variable
Average
Regression coefficients (t-statistics)d)
Model I
Constant (Euro/ha UAA)
182.90
Share of grassland in % UAA b)
31.43
East Germany (yes = 1; no = 0)
0.26
Precipitation in March - June in mm c)
East Germany and precipitation in
March - June in mm c)
Mean annual temperature in °C c)
120.065 (2.14)
-1.735 (-10.24)
-192.154 (-3.35)
Model II
74.588 (0.32)
-218.780 (-3.39)
270.17
-
-0.397 (-4.14)
55.81
0.465 (1.95)
0.531 (2.01)
8.40
16.532 (2.71)
27.736 (4.08)
λ (coefficient reflecting spatial error autocorrelation)
adjusted R2
0.834 (31.64)
80.87 %
0.838 (31.95)
77.23 %
a)
Mean annual land rental price on district level in 1999, according to Statistisches Landesamt
Baden-Württemberg (2007). - b) Data from agricultural census 1999 according to Statistische
Ämter des Bundes und der Länder (2001). - c) Own calculations based on DWD (2007). - d) n
= 439; in brackets: asymptotic t-statistics; except the constant in model II all coefficients are
significant at a 5% level.
Source: Lippert et al., 2009
The results show a significant correlation between land rents and increasing mean
temperatures, as well as (except for East Germany) a significant increase of land rents along
with declining spring precipitation. For East Germany the relation between spring
precipitation (March to June) and land rents is however significantly positive. Further, East
Germany (i.e. the new federal states) has a significantly lower level of land rental prices.
The differences between the models arise from the explicit inclusion of grassland in model I:
its share significantly correlates both with spring precipitation (correlation coefficient =
0.61 ***) and with annual average temperature (correlation coefficient = -0.34 ***). Further,
grassland shares correlate negatively with land rents. This means that at least as long as it is
allowed for farmers to convert grassland to arable land both a reduction in spring precipitation
as well as a temperature increase will lead to a reduction of grassland which results in the
stronger marginal effects of temperature and precipitation on land rents in model II .
Which of the two models is appropriate depends on the future legal possibilities of farmers to
convert grassland in reaction to climate change. At present such a conversion is limited
because of the cross compliance regulations of the Common Agricultural Policy (CAP).
However, in a decade or two this restriction could be lifted due to increasing food prices. On
the other hand, in the interpretation of the results one has to consider that natural site
conditions, especially slope may hamper in some cases a change to arable land use.
4
Predictions
Next, we used average predicted future climate data in order to calculate local land rent
changes under different IPCC scenarios for a time horizon between 2011 and 2040. The
5
corresponding IPCC (2007) scenario data was taken from the regional climate model REMO
(Jacob, 2001; UBA, 2006). REMO is based on a former weather prediction model of the
German Weather Service (Europamodell) which was improved by implementing physical
parameterisations of the global climate model ECHAM3/T42 and ECHAM4/T106 (cf. Jacob
and Podzun, 1997; Jacob, 2001).
When the average climate data from 2011 to 2040 according to REMO are inserted into the
above estimation equations, for different IPCC scenarios the estimated land rents as given in
Table 2 are obtained. With an average net income of German agriculture in the years 1998 to
2000 of 10,566 million Euros (BMVEL, 2004, p. 127) the accumulated increase of land rents
of 600 million Euros accounts for 5 to 6 % of net German agricultural income. The weighted
average increase is between 31 and 36 Euros per hectare UAA, which corresponds to an
increase by 17 % to 20 % of the average land rent in the reference situation. Figure 3 shows
the relatively high variability of the estimated land rent increases for the IPCC scenario A1B.
Table 2: Estimated future climate and resulting land rents
Scenario a)
Period
Spring precipitation (March - June) in mm b)
East Germany and spring precipitation in mm b)
Average annual temperature in °C b)
Model I: Change in land rents (Mio. €) c)
weighted average change (€/ha UAA) d)
Model II: Change in land rents (Mio. €) c)
weighted average change (€/ha UAA) d)
a)
Reference
1961-‘90
270.17
55.81
8.40
A1B
2011-‘40
314.18
70.83
9.86
568.59
33.14
611.90
35.66
A2
2011-‘40
325.54
71.36
9.97
599.17
34.92
597.81
34.84
B1
2011-‘40
302.02
67.41
9.82
527.21
30.73
623.14
36.32
IPCC scenarios. - b) Calculations based on DWD (2007) and MPI on behalf of the
Umweltbundesamt (2006) (REMO-Model). - c) Total change for Germany = ∑i UAAi times
(difference between estimated land rental prices after climate change and estimated prices for
1999)i for the 439 German districts (i = 1, …, 439); for model I and II cf. Table 1. d)
Weighted by the utilisable agricultural area (UAAi) of the districts.
Source: Lippert et al., 2009
6
Figure 3: Change of the land rental prices for scenario A1B according to model I (left) and
model II (right)
Depiction from Lippert et al. (2009)
According to Figure 3 in West Germany the positive land rent changes are predicted to rise
from the north to the south. This is a result of higher temperature increases in Southern
Germany. The estimated greater increase in East German land rents is due to the presently
more unfavourable climatic water balance in this part of the country which is supposed to
profit from the predicted local increases in spring precipitation.
5
Discussion and scope for future research
The above estimations of future land rent changes were done for moderate temperature and
precipitation changes. Anyway, estimations should only be based on predicted climate
variables that do not exceed the observed range of these variables in the cross-sectional
analysis. In case of strong climatic change leading to local climates that have not been
observed in any German district so far the regression equations found should not be used.
Albeit, it cannot be excluded that in the long run, when the changes in the climatic conditions
will be more severe than those simulated in this study, the net impact on German agricultural
income will be negative.
Moreover, it should be noted that the predicted beneficial effects of climate change on
German agriculture may be partly offset by higher yield variability due to more frequent
extreme precipitations and droughts which, in the end, would result in increasing insurance
cost. Also, new pests and animal diseases which are unknown so far in Central Europe may
appear as a consequence of climate change.
On the other hand, in our analysis we implicitly assumed a constant price level for agricultural
produce. However, because of the negative consequences of climate change in other parts of
the world, world prices will tend to increase (cf. von Braun, 2007, p. 3f). This would lead to a
7
further advantage for a German agriculture with a higher yield potential when compared to
the present situation.
In case of omitted explaining variables that correlate with climate variables used in the
regression equations marginal climate effects may be overestimated. This problem is
conceivable for the estimated effect of average slope of local farmland: supposing that
farmland at higher altitudes is simultaneously characterized by lower mean temperatures and
strongly sloped agricultural area (which usually is less profitable) only part of the lower
farmland profitability at higher altitudes would result from reduced temperatures.
Due to the comparative-static character of the Ricardian approach a further aspect we did not
account for when estimating land rent changes are climate related adaptation costs in the strict
sense.
In the future we plan to extend and improve our research especially regarding the two latter
aspects:
(a) When an omitted variable correlates with one of the land value explaining climate
variables used in the Ricardian analysis the results will be biased. For instance, if
temperature was correlated with soil quality this would lead to a substantial
overestimation of the effect of the former variable on the land rent. Consequently, we
will try to assess the effects of different ways to account for soil quality by comparing
the results of several feasible approaches to control for German districts’ average soil
attributes. Extrapolating broad soil types from the FAO soil classification (as done by
Seo and Mendelsohn, 2008a, p. 111; 2008b) will be compared to the use of the more
detailed German soil map (cf. BGR, 2009) and to the use of the sophisticated German
soil index. For selected regions of Southern Germany we will also try to do
corresponding analyses at lower spatial scales than district level. Such comparisons
will give interesting insights into the reliability of common Ricardian analyses at
higher spatial scales. Here, an important aim will be to find out whether a coarse
consideration, e. g. by using dummies for main soil types, has a strong impact on the
outcome.
(b) So far, common Ricardian analysis is just a “black box” approach which does not
reveal how climate change affects farmers’ adaptation decisions and the resulting
pattern of cultivated crops and farming activities. Thus, it seems to be promising to
apply and to further develop the multinominal logit approach recently suggested by
Seo and Mendelsohn (2008a, b) at a lower spatial scale. For such an approach
observed crop shares but also reported main farm types may be used to describe
existing farming patterns. The objective is to explain the climate dependency of
German land use patterns and systems in order to get a better idea on how farmers’
adaptation measures may look like in future. Next, based on theoretical considerations
on climate change adaptation options and related adjustment cost (cf. Smit and
Skinner, 2002; Kelly et al., 2005) we want to develop an appropriate typology of all
relevant transition cost to be borne by the farming sector. Using this typology,
regional adjustment difficulties related to the predicted changes may be identified
accounting for socioeconomic as well as ecological constraints in the different
German districts. Thus, we want to elaborate a systematic way to detect which climate
change induced agricultural adaptation needs and related costs different regions will
8
References
Anselin L. (1988). Spatial Econometrics: Methods and Models. Kluwer, Dordrecht.
BGR (Bundesanstalt für Geowissenschaften und Rohstoffe) (2009). Information on
Bodenübersichtskarte der Bundesrepublik Deutschland 1:1.000.000 (BÜK1000) in
http://www.bgr.bund.de/cln_092/nn_325378/DE/Themen/Boden/Produkte/Karten/BU
EK__1000.html (accessed on July 17th, 2009).
BMVEL (Bundesministerium für Verbraucherschutz, Ernährung und Landwirtschaft) (2004).
Ernährungs- und agrarpolitischer Bericht der Bundesregierung 2004. Berlin, Germany.
DWD (Deutscher Wetterdienst) (2007). Mittelwerte aller DWD Stationen der Periode 19611990, www.dwd.de/de/FundE/Klima/KLIS/daten/online/nat/index_mittelwerte.htm.
Accessed 31 July 2007.
IPCC (Intergovernmental Panel on Climate Change) (2007). Summary for Policymakers. In:
Solomon S, Qin D, Manning M, Chen Z, Marquis M, Averyt KB, Tignor M, Miller
HL (eds) Climate Change 2007: The Physical Science Basis. Contribution of Working
Group I to the Fourth Assessment Report of the Intergovernmental Panel of Climate
Change. Cambridge University Press, Cambridge.
Jacob D. (2001). A note to the simulation of the annual and inter-annual variability of the
water budget over the Baltic Sea drainage basin. Meteorol Atmos Phys 77, 61-73.
Jacob D., Podzun, R. (1997). Sensitivity studies with the regional climate model REMO.
Meteorol Atmos Phys 63, 119-129.
Kelly D. L., Kolstad C. D., Mitchell G. T. (2005). Adjustment costs from environmental
change. Journal of Environmental Economics and Management 50, 468-495.
LeSage J. (1999). Spatial Econometrics. Department of Economics, University of Toledo
(USA), www.rri.wvu.edu/WebBook/LeSage/spatial/spatial.html. Accessed 4 Feb
2003.
Lippert C., Krimly T., Aurbacher J. (2009). A Ricardian analysis of the impact of climate
change on agriculture in Germany. Climatic Change 97 (3/4), 593-610 (DOI
10.1007/s10584-009-9652-9).
Mendelsohn R., Nordhaus W. D., Shaw D. (1994). The Impact of Global Warming on
Agriculture: A Ricardian Analysis. Am Econ Rev 84(4), 753-771.
MPI on behalf of the Umweltbundesamt (2006). REMO-UBA scenario data. CERA-database
(Climate and Environmental Data Retrieval and Archive), WDCC (World Data Center
for Climate). Modelle & Daten. http://cera-www.dkrz.de. Accessed 27 Nov 2007.
9
Seo N., Mendelsohn R. (2008a). An analysis of crop choice: Adapting to climate change in
South American farms. Ecological Economics 67, 109-116.
Seo N., Mendelsohn R. (2008b). Measuring impacts and adaptations to climate change: a
structural Ricardian model of African livestock management. Agricultural Economics 38,
151-165.
Smit B., Skinner M. W. (2002). Adaptation Options in Agriculture to Climate Change: a
Typology. Mitigation and Adaptation Strategies for Global Change 7, 85-114.
Statistische Ämter des Bundes und der Länder (2001). Statistik regional. Daten und
Informationen. CD ROM.
Statistisches Landesamt Baden-Württemberg (2007). Communication of the 1999 land rental
prices in German districts, collected by the statistical offices of Germany. Stuttgart,
August 2007.
von Braun J. (2007). The World Food Situation: New Driving Forces and Required Actions.
IFPRI’s Biannual Overview of the World Food Situation presented to the CGIAR
Annual General Meeting, Beijing, December 4, 2007. International Food Policy
Research Institute (IFPRI), Washington.
UBA (Umweltbundesamt) (2006). Künftige Klimaänderungen in Deutschland - Regionale
Projektionen für das 21. Jahrhundert. Hintergrundpapier, April 2006, updated in
September 2006. www.umweltbundesamt.de.
Zebisch M., Grothman T., Schröter D., Hasse C., Fritsch U., Cramer W. (2005). Climate
Change in Germany, Vulnerability and adaptation of climate sensitive sectors.
Environmental research of the federal ministry of the environment, nature
conservation and nuclear safety, research report 201 41 253, UBA-FB 000844/e.
www.umweltbundesamt.de.
10